Integrating Patient- reported Measures and Predictive Data ......JAMA 2012; 308(12): 1217-18 ....
Transcript of Integrating Patient- reported Measures and Predictive Data ......JAMA 2012; 308(12): 1217-18 ....
Integrating Patient-reported Measures and Predictive Data in a Shared Decision Report for
Knee and Hip Arthritis Treatment Decisions
Hua Zheng, PhD1, Danielle Lavallee, PharmD PhD2, Christina Bond, MS3, Patricia Franklin, MD MBA MPH4
1University of Massachusetts Medical School2University of Washington Medical Center
3Patient Research Group4Northwestern University Feinberg School of Medicine
Funding
• PCORI This research was funded through a Patient-Centered Outcomes Research Institute (PCORI) Award IHS-1507-31714 (Franklin).
The views in this work are solely the responsibility of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute (PCORI), its Board of Governors or Methodology Committee.
• AHRQ FORCE-TJR was funded by a grant from the Agency for Healthcare Research and Quality P50HS018910 (Franklin, 2010-2015).
PROs for shared decision making
Shared decision making is a key component of patient-centered health care. Patients and clinicians working together to select treatments and care plans can improve patient outcomes.
The use of patient-reported outcomes (PROs), in addition to EHR data, to facilitate patient-clinician shared decisions is new and poses new informatics challenges.
EHR: strengths and limitations in shared decision tools
EHR Limitations
The electronic health record (EHR) digitally aggregates patient’s health data at each episode. EHR systems:• Capture and store clinical data per episode for medical, legal,
and administrative purposes.
• EHRs collect data locally. If the patient visits different hospitals or clinicians, today’s EHR does not integrate patients’ health data across clinical settings and time.
• EHRs collect data when the patient visits a hospital or MD, and may miss data if the patient does not visit; thus, data are potentially incomplete and biased.
• Decision tools within EHRs are based on local data, not national data.
• Develop and test a novel shared decision tool (with patient-centered predictive analytics) for surgeons and patients with knee/hip osteoarthritis (OA) who are considering total joint replacement (TJR) surgery.
• Evaluate if this tailored shared decision tool will (a) enhance OA patient's decision quality, (b) influence OA outcomes after medical or surgical treatment.
A.S.K. Study: Arthritis care through Shared Knowledge
A.S.K. Study based on FORCE-TJR RegistryFORCE-TJR: Function and Outcomes Research for Comparative Effectiveness in Total Joint ReplacementNational TJR research cohort/registry
Primary outcome = Patient-reported pain and functional (plus clinical data)
6
JAMA 2012; 308(12): 1217-18
FORCE-TJR: US representative sample of200 surgeons; 30,000 patients in 28 States
• Stratified random sample of US surgeons
• 75% are community-based• Fellowship-trained and general
orthopedic surgeons• High and low volume
surgeons/hospitals; urban and rural• Teaching hospitals, non-teaching• Private, public and HMO insurance• All major implant manufacturers
National norms from 2011-2016Medicare and Private Insurers
FORCE-TJR: key capabilities to build shared decision tools
• Collect data from patients directly (not limited to one EHR), but can return data to patient or EHR for storage
• Data capture process is pre-determined and scheduled, so does not miss follow-up time points
• Complete, pre-defined risk and outcome data as well as PRO data
• Capture data from representative patients across country, so can provide national norms and support access to national databases
Patient-reported symptoms and risks compared with national norms
• Collect PRO and symptom data from patients directly
• Data capture process is pre-determined and scheduled, so will not miss follow-up time points
• Obtain needed risk and outcome data from patient survey
• Capture data from OA patients across country, so can provide national norms and support access to national databases
Trended Symptomsover Time
Patient Risk Profile
Color-coded Norms
Predictive Analytics: What outcomes are patients like me likely to achieve?
• Collect data from patients directly
• Data capture process is pre-determined and scheduled, so will not miss follow-up time points
• Obtain needed risk and outcome data from patient survey
• Capture data from patients across country, so can provide national norms and support access to national databases
Individualized predictive outcomes based on
similar patients
Multi-Variate Models for Outcome Prediction
Multi-variable risk models for key outcomes
A.S.K. predictive shared-decision reports generated real-time for patient and clinician
Non-operative comparativeinformation
Patient Feedback on Report Design
“A graph might work. I’m a visual person…. to see the little stacks (bar graphs), like you came in here a year ago and you were at 20 and one year later you are at 70 and you’re working towards a goal of 80.”
“In my mind [the report] is part of the treatment plan. A lot of times they leave the patient out. They give the diagnosis and tell them to take that pill or whatever. By giving the answers to the questions, it helps you be involved in the actual decision making and to understand why.”
Clinician Feedback on Report Design
Color-coding definitions and
norms
Wording and content modifications
MD (CT): “The report is great for highlighting the risks to patients, and managing expectations.”
Hospital IT Review/Approvals Require Time; All Approved A.S.K.
A.S.K.Team
Site IT Department
Web-based data capture and
reporting is newto clinics.
One-page IT overview
Site IT Questionnaire
Meeting to review detailed documents
Approval
Return Questionnaire; other documents
*Process in takes time.Orange
Schedule r review
Site Review
Site Review
Completing Questionnaire
FORCE National Registry Database
A.S.K.: Integrate Patient, National, EHR Data
1. Survey Request
2. Survey Return
Real-Time Generated ReportIndividual
Survey Data
Update Registry Data
National Norms
Exam Room
EHR System
3. Shared Decision Report
4. Update Norms and
Models
Statistical Models
5. API(Future)
WEB-BASED
A.S.K. Shared decision report based on patient-reported data
1. Trended patient-reported symptoms/PROs- Pain, activities of daily living, physical function
2. Risk factors for post-TJR complications3. Predicted TJR outcomes for similar patients
- Pain relief, functional gains
4. Predicted TJR impact on pain with patients’ priority activities5. Complication risks6. Summary evidence for non-surgical treatments
Technical Lessons Learned• Returning data to EHRs
• Longitudinal registry data collected in parallel can supplement EHR data.
• Returning data to EHRs may involve processes like identifier matching, new field creation, automated and secure data transfer, etc.
• Security and privacy review by hospitals/practices• To protect patients’ data, security and privacy checks are
required by all sites. The review and approval process varies and ranges from 1-2 months to 4-5 months.
• Research IT implementation is a new concept.
• FORCE-TJR web system can collect patient-reported data as a third-party platform, and readily interface with EHRs (e.g., API) for storage of the PRO data and decision reports.
• Discrete data and/or reports can be emailed to patients and additional clinicians (e.g., primary care doctors)-even if not using same EHR.
• Longitudinal direct-to-patient data capture is automated and can support EHRs to store complete “picture” of treatment impact, disease progression.
Dissemination and EHR Inter-operability
Thank You!
A.S.K. [email protected]